An analysis of perceptual errors in perspective displays

Abstract

Display dimensionality 1s one of the most debated issues m the design of cockpit-based displays of air traffic information. Many practitioners agree that presenting air traffic information on an integrated 30 display is preferable, when compared to presenting it on a 20 planar or co-planar display. However, research has shown that operators make errors in estimating the location of objects in 30 displays. This may be because of perceptual distortions caused by the geometric parameters used to generate the image. However, despite the issues identified regarding the locating of objects, some studies have found other performance advantages associated with these 30 displays. Therefore, it seems that attempts should be made to minimise perceptual biases so that these displays can be utilised to present integrated information in 30 environments. The aim of this thesis was to develop a model of distance estimation errors in perspective displays. It was hypothesised that many of the perceptual errors observed in perspective displays, such as azimuth and inter-object distance estimation errors, were related to observers wrongly estimating the distance between themselves and objects in the virtual world. Four experiments examining inter-object distance estimation were conducted. Participants were required to set a perspective image of a box to represent a perfect cube (requiring them to make a distance estimation scaled relative to the frontoparallel plane). Results showed that participants made inter-object distance estimation errors that increased as the distance between the observer and the objects in the display increased. Based on these results, two models explaining interobject distance estimation errors were developed. The first model postulated that participants underestimated the distance between themselves and objects in the display. The second model suggested that participants used 20 ( on-screen) cues to set the box to a cube. These models were applied to azimuth estimation errors observed in studies of perspective displays. It was found that while azimuth estimation error could only be partially modelled as a distance perception error, it could be explained to a greater extent by applying a strategy based on the 20 (on-screen) image. The findings of this study indicated that either distance estimation errors or 20 strategies could account for inter-object distance estimation errors in perspective displays.